Ethernet switches are growing in capability. As a consequence the role of Ethernet is rapidly expanding in networks that were the domain of other technologies such as SONET/SDH TDM and ATM. The question of how Ethernet will evolve and what capabilities it can offer in these areas is still under development.
Ethernet as specified today is a system. How spanning tree, data plane flooding and MAC learning combine to populate forwarding tables and produce resilient any-to-any behavior in a bridged network is well understood. What is less obvious is that the resulting behavior is purely a consequence of this particular combination of functions combined with what the underlying hardware can do, and that by simply disabling some Ethernet functionality, it is possible to employ alternative control planes and obtain different forwarding behaviors.
It is desirable to be able to drive Ethernet towards increasingly deterministic behavior. One behavior of note is that of Provider Backbone Transport (PBT) as disclosed in commonly assigned U.S patent application no. US20050220096 filed Apr. 4, 2004 and hereby incorporated by reference. Using PBT, Ethernet switches may perform PBT MAC forwarding on the basis of a statically configured VID/MAC tuple. This means the forwarding hardware performs a full 60 bit lookup (VID(12)+MAC DA (48)) only requiring uniqueness of the full 60 bits for forwarding to resolve correctly.
Generalized Multi-protocol Label Switching (GMPLS) extends MPLS to provide the control plane (signaling and routing) for devices that switch in any of these domains: packet, time, wavelength, and fiber. GMPLS signaling is well suited to setup paths with labels but it does require a minimal IP control plane and IP connectivity so it is suited to certain scenarios where a large number of paths or dynamic path management is required. The common control plane promises to simplify network operation and management by automating end-to-end provisioning of connections, managing network resources, and providing the level of QoS that is expected in the new, sophisticated applications.
Ethernet provider backbone transport (PBT) paths are controlled utilizing Generalized Multi-protocol Label Switching (GMPLS) signaling protocol as described in commonly assigned U.S. patent application Ser. No. 11/580,796 filed on Oct. 13, 2006. A path between edge nodes is identified by a combination of a VID and destination MAC address in a VID/MAC tuple populated in the forwarding tables of intermediary nodes. To establish the PBT path, a path calculation is performed from the originator node to the terminator node through the network. The originating node then sends a GMPLS label object with a preferred VID/MAC to identify the path to the originator. The intermediary nodes or bridges forward the object to the terminating node. The terminating node then offers a VID/MAC tuple in a GMPLS label object for the path to the terminating node in response. When the intermediary nodes forward the response from the terminating node to the originator, the appropriate forwarding labels are then installed in the forwarding tables of each node utilizing the associated VID/MAC tuples.
As Ethernet expands into provider network these is a need to leverage the benefits of GMPLS deterministic behavior with the flexibility of Ethernet. MPLS signaling has recently been extended to include P2MP and GMPLS could be trivially extended to take advantage of this Point-to-multipoint (P2MP) trees mapped though Ethernet networks can be vulnerable to network failures. In the MPLS world such failures are normally handled via either local repair techniques or control plane restoration. However, in the case of P2MP trees mapped though Ethernet, the normal operation of conventional resiliency schemes can result in highly undesirable results (and such issues are not exclusively confined to Ethernet). A single failure may impact a significant number of leaves on the tree for which restoration becomes a non-trivial undertaking. Reconstructing a completely new P2MP tree from scratch is not an option due to the time and resources required. Attempts to incrementally modify a live tree to circumvent a network failure may introduce loops and potentially leave forwarding artifacts. Restoration of a tree requires incrementally signaling new connectivity to each leaf which may take a large amount of time. Further the computation required for optimal trees may be non-trivial (e.g. Steiner tree which is an NP-hard problem), suggesting pre-planned resilience options are required.
Accordingly, methods and apparatus enabling Point-to-multipoint (P2MP) resilience for GMPLS control of Ethernet remain highly desirable.